A Comparison of Preference-Based, Generic and Disease-Specific Health-Related Quality of Life in Pediatric Inflammatory Bowel Disease

Abstract

Objective

Generic preference-based HRQOL assessments used expressly for economic evaluation have not been examined in pediatric Crohn’s disease (CD) and ulcerative colitis (UC). The objective was to further assess the construct validity of preference-based HRQOL measures in pediatric IBD by comparing the Child Health Utility 9 Dimensions (CHU9D) and Health Utilities Index (HUI) to the disease-specific IMPACT-III and to the generic PedsQL in children with CD and with UC.

Methods

The CHU9D, HUI, IMPACT-III and/or PedsQL were administered to Canadian children aged 6 to 18 years with CD and UC. CHU9D total and domain utilities were calculated using adult and youth tariffs. HUI total and attribute utilities were determined for the HUI2 and HUI3. Total scores for IMPACT-III and PedsQL were determined. Spearman correlations were calculated between generic preference-based utilities and the IMPACT-III and PedsQL scores.

Results

The questionnaires were administered to 157 children with CD and 73 children with UC. Moderate to strong correlations were observed between the CHU9D, HUI2, HUI3 and the disease-specific IMPACT-III or generic PedsQL. As hypothesized, domains with similar constructs demonstrated stronger correlations, such as the Pain and Well-being domains.

Conclusions

While all questionnaires were moderately correlated with the IMPACT-III and PedsQL questionnaires, the CHU9D using youth tariffs and the HUI3 were most strongly correlated and would be suitable choices to generate health utilities for children with CD or UC for the purpose of economic evaluation of treatments in pediatric IBD.

INTRODUCTION

Inflammatory bowel disease (IBD) which includes ulcerative colitis (UC) and Crohn’s disease (CD), is characterized by unpredictable flares of gastrointestinal inflammation, abdominal pain, diarrhoea, fatigue and weight loss which and periods of quiescence which can affect health-related quality of life (HRQOL) (1). While pediatric IBD disease-specific HRQOL instruments are useful measures of quality of life impacts (2,3), economic evaluations that evaluate emerging treatments for IBD require generic preference-based instruments that generate utilities, i.e., preferences, for disease heath states. Generic preference-based HRQOL instruments allow for the determination of quality-adjusted life years (QALYs) whereby utilities—a measure of morbidity—are multiplied by life expectancy. QALYs are a powerful universal outcome to enable comparison of the cost-effectiveness of IBD treatments to interventions in other diseases to facilitate budget allocation decision-making (4). Due to the lack of such tools in child health, pediatric economic evaluations have relied on utilities generated from adult studies.

Currently, there is no gold standard generic preference-based HRQOL instrument for children. The Child Health Utility 9 Dimensions (CHU9D) was developed as a generic, preference-based HRQOL measure to reflect health states relevant to children (5,6). Earlier in this research, the CHU9D was found to be valid and reliable in pediatric CD and UC (7). The objective of the study was to further assess the validity of the CHU9D and the Health Utility Index (HUI) in pediatric CD and UC by comparing them to disease-specific and generic HRQOL instruments used in pediatric IBD. The comparison can inform the selection of appropriate HRQOL tools for future cost-effectiveness analysis.

METHODS

The study was approved by the Research Ethics Boards of the Hospital for Sick Children (#1000039604), the IWK Health Centre (#1015558) and the Janeway Children’s Health and Rehabilitation Centre (#13.229). All parent participants provided informed consent and children provided informed assent.

Study Design and Participants

Participants were recruited from three study sites of the Canadian Children Inflammatory Bowel Disease Network (CIDsCaNN) (https://cidscann.ca), a longitudinal repeated measures observational cohort of Canadian children with newly diagnosed CD and UC. Children aged 6 to 18 years without co-morbidities were recruited between February 2014 and December 2018. While children may have been recruited at a young age, the HRQOL questionnaires were only administered after the child reached the age at which the questionnaire had been validated.

MEASURES

The CHU9D and HUI are indirect preference-based HRQOL instrument that consist of a classification system which is administered as a questionnaire, and an underlying valuation algorithm or tariff set that includes previously derived utilities for each level of each attribute/domain. The CHU9D covers nine dimensions: Worried, Sad, Pain, Tired, Annoyed, Schoolwork, Sleep, Daily routine and Activities, and each dimension is represented by five levels such as Not, A little bit, A bit, Quite and Very (5,6,8). CHU9D utility weights were calculated using both the Australian adult tariffs and Australian adolescent tariffs (9). The HUI2 classification system includes seven attributes: Sensation, Mobility, Emotion, Cognition, Self-Care, Pain and Fertility, each with three to five levels (10). The fertility attribute was omitted for the purpose of this study. The HUI3 classification system includes eight attributes: Vision, Hearing, Speech, Ambulation, Dexterity, Emotion, Cognition and Pain, each with five or six levels (10). The Pediatric Quality of Life Inventory version 4.0 Generic Core Scales (PedsQL) HRQOL questionnaire was developed and validated to assess quality of life in children aged 2 to 18 years (11). The PedsQL has four domains: Physical Functioning, Emotional Functioning, Social Functioning and School Functioning. PedsQL scoring was conducted according to the algorithm related to the age-range-specific questionnaire completed by each participant. The disease-specific IMPACT-III includes four domains: Well-being, Social functioning, Emotional functioning and Body Image (12). Disease activity was assessed using the Pediatric Crohn’s Disease Activity Index (wPCDAI) (13) in CD and the Pediatric Ulcerative Colitis Activity Index (PUCAI) (14) in UC at the time of HRQOL assessment and was categorized as quiescent, mild, moderate, or severe. Additional details are provided in the supplementary material.

Questionnaire Administration

Pediatric patients with IBD were administered the CHU9D, HUI, IMPACT-III and PedsQL questionnaires in addition to other CIDsCaNN measures at repeated assessments. There was no fixed schedule of repeat assessment; questionnaires were administered at various intervals at the discretion of the clinician. All questionnaires were administered electronically in English via REDCap and were self-completed by children or with interviewer assistance. If needed, interviewers assisted by reading the questions to respondents and with physical completion of the questionnaire with the child’s responses. Interviewers were not permitted to prompt responses. Results from the first date that all questionnaires were completed on the same day were scored.

Statistical Analysis

Data were analyzed using R v. 4.0.0 (15) statistical packages. CD and UC data were analyzed separately. Chi-square, t-tests and Kruskal-Wallis rank-sum tests were used for demographic comparisons between sexes. CHU9D, HUI2 and HUI3 health utilities data were not normally distributed according to the Shapiro–Wilk normality test. A normal distribution was also rejected for the IMPACT-III and PedsQL scores (P < 0.05) thus non-parametric statistical tests including the Kruskal–Wallis rank-sum test was used to compare instruments.

The CHU9D, HUI, IMPACTIII and PedsQL scoring algorithms were coded in R. Incomplete questionnaires were excluded. A priori hypotheses regarding expected correlations between domains and attributes measuring similar constructs were postulated. It was hypothesized that correlations >0.5 would be observed for conceptually similar domains and correlations between 0.35 and 0.5 for overlapping but conceptually distinct domains (Table 1). As the data were not normally distributed, Spearman correlations were calculated between: CHU9D with youth or adult tariffs and IMPACT-III; HUI2 and IMPACT-III; HUI3 and IMPACT-III; CHU9D with youth or adult tariffs and PedsQL; HUI2 and PedsQL; and HUI3 and PedsQL. Domain correlations between CHU9D and HUI have been previously reported (7). Domain correlations for the HUI3 Dexterity, Vision, Hearing and Speech domains or the HUI2 Sensation domain were not tested as similar domains were not apparent in the PedsQL and IMPACT-III.

Table 1.

Estimated a priori correlations between domains of quality-of-life questionnaires in children with inflammatory bowel disease

		IMPACT-III Domains				PEDQL Domains
	Domain	Well-Being	Emotional Functioning	Social Functioning	Body Image	Physical Functioning	Emotional Functioning	Social Functioning	School Functioning
CHU9D	Worry		>0.5	0.35–0.5	0.35–0.5		>0.5	0.35–0.5	0.35–0.5
	Sad		>0.5	0.35–0.5	0.35–0.5		>0.5	0.35–0.5
	Pain	>0.5				>0.5
	Tired	0.35–0.5		0.35–0.5		0.35–0.5		>0.5	0.35–0.5
	Annoyed		>0.5				>0.5	0.35–0.5	0.35–0.5
	School-work			0.35–0.5					>0.5
	Sleep	0.35–0.5						0.35–0.5	0.35–0.5
	Daily Routine	0.35–0.5		0.35–0.5		>0.5		0.35–0.5	0.35–0.5
	Activities	0.35–0.5		0.35–0.5		>0.5		0.35–0.5	0.35–0.5
HUI2	Emotion		>0.5		0.35–0.5		>0.5
	Cognition			0.35–0.5				0.35–0.5	>0.5
	Mobility	0.35–0.5				>0.5
	Self-care	0.35–0.5				>0.5
	Pain	>0.5				>0.5
HUI3	Emotion		>0.5		0.35–0.5		>0.5
	Cognition			>0.5				0.35–0.5	>0.5
	Ambulation					0.35–0.5
	Pain		>0.5			>0.5

RESULTS

Participant Characteristics

A total of 230 children participated in one or more comparisons between the preference-based measures and the IMPACT-III or PedsQL of whom 157 (68.3%) were diagnosed with CD and 73 (31.7%) with UC. Characteristics for the sub-sample completing preference-based measures and the IMPACT-III (94.3% of total) are presented in Table 2. In CD, 38.5% of participants were female and in UC, 56.1% were female. The mean age was 14 years. Sixty-four percent of participants were of Caucasian origin. Participants were predominantly experiencing quiescent or mild disease activity. Demographics of the n = 129 participants completing the CHU9D, HUI and PedsQL were similar (results not shown). In CD, 67 participants were included in the comparison between preference-based measures and both IMPACT-III and PedsQL. In UC, 46 participants were included in the comparison between preference-based measures and both IMPACT-III and PedsQL. Correlations of preference-based measures with IMPACT-III and with PedsQL were analyzed separately because the timing of questionnaire completion did not coincide for all participants.

Table 2.

Patient characteristics of subjects with CD and UC that completed at least one date-matched CHU-9D, HUI and PedsQL assessments

Characteristic	CD	UC	Total
N	148	66	214
Sex at birth female (%)	57 (38.5)	37 (56.1)	94 (43.9)
Mean age, years (SD)	13.9 (2.3)	13.9 (2.7)	13.9 (2.4)
Age range, years	7.6–18.0	6.7–18.1	6.7–18.1
Ethnicity (%)
Caucasian	95 (64.2)	45 (68.2)	140 (65.4)
Mixed	17 (11.5)	2 (0.3)	19 (8.9)
South Asian	9 (6.1)	11 (16.7)	20 (9.3)
East and Southeast Asian	5 (3.4)	1 (1.5)	6 (2.8)
Caribbean	3 (2.0)	1 (1.5)	4 (1.9)
West Central Asian and Middle Eastern	3 (2.0)	3 (4.5)	6 (2.8)
Latin, Central and South American	1 (0.7)	0 (0.0)	1 (0.5)
North, South, Central, East and West African	7 (4.7)	2 (3.0)	9 (4.2)
Other or Unknown	8 (5.4)	1 (1.5)	9 (4.2)
Clinical Site (%)
Toronto, Ontario	105 (70.9)	55 (83.3)	160 (74.8)
Halifax, Nova Scotia	42 (28.4)	10 (15.2)	52 (24.3)
St. John’s, Newfoundland	1 (0.7)	1 (1.5)	2 (0.9)
Disease activity category *
Quiescent	55 (37.2)	37 (56.0)	92 (43.0)
Mild	42 (28.4)	13 (19.7)	55 (25.7)
Moderate	16 (10.8)	6 (9.1)	22 (10.3)
Severe	20 (13.5)	8 (12.1)	28 (13.1)
Missing/unknown	15 (10.1)	2 (3.0)	17 (7.9)

^*Disease activity measured with weighted Pediatric Crohn’s Disease Activity Index for CD and with Pediatric Ulcerative Colitis Activity Index for UC. CD = Crohn’s disease, UC = ulcerative colitis.

Correlation Between Preference-Based and Non-Preference-Based HRQOL Measures in CD

Spearman correlations between overall scores of the CHU9D, HUI, IMPACT-III and PedsQL in CD and UC are shown in Table 3. As correlations between CHU9D with adult tariffs and IMPACT-III and PedsQL were similar to those found with youth tariffs, only the latter findings are presented. In CD, The CHU9D and HUI showed similar moderate statistically significant correlations with IMPACT-III ranging from 0.52 to 0.61 across questionnaires and tariff sets. The CHU9D and HUI showed very similar moderate-to-strong statistically significant correlations with PedsQL (range 0.67 to 0.71) regardless of questionnaire or tariff set.

Table 3.

Spearman correlations between overall scores of generic preference-based questionnaires and the IMPACT-III or PedsQL questionnaires

	IMPACT III		PedsQL
	CD	UC	CD	UC
Number of paired assessments	76	53	148	66
CHU-9D adult tariffs	0.52*	0.79*	0.67*	0.71*
CHU-9D with youth tariffs	0.58*	0.81*	0.69*	0.69*
HUI2	0.61*	0.67*	0.67*	0.51*
HUI3	0.55*	0.70*	0.71*	0.67*

^* P < 0.0001. CD = Crohn’s disease; UC = ulcerative colitis.

Spearman correlations between the CHU9D, HUI2 and HUI3 domains and IMPACT III and PedsQL domains are shown in Table 4. As hypothesized, in CD, CHU9D Pain was moderate-to-strongly correlated (r = 0.58, P < 0.05) with IMPACT-III Well-being. In addition, CHU9D Activities was moderately correlated to IMPACT-III Well-being (r = 0.50, P < 0.05). With regard to HUI, HUI2 Pain was moderate-to-strongly correlated with the IMPACT-III Well-being (r = 0.64, P < 0.05). HUI3 Pain was moderate-to-strongly correlated with IMPACT-III Well-being (r = 0.71, P < 0.05). The correlation between HUI3 Pain and IMPACT-III Well-being was higher than similar HUI2 and IMPACT-III domains. HUI3 Emotion was moderately correlated with the IMPACT-III Body Image (r = 0.39, P < 0.05).

Table 4.

Spearman correlations between questionnaire domains in CD and UC

			IMPACT-III Domains				PEDQL Domains
	Domain		Well-being	Emotional Functioning	Social Functioning	Body Image	Physical Functioning	Emotional Functioning	Social Functioning	School Functioning
CHU9D youth tariffs	Worry	CD		0.42	0.45	0.27		0.57	0.41	0.34
		UC		0.54	0.63	0.32		0.49	0.24	0.06*
	Sad	CD		0.12*	0.16*	0.06*		0.41	0.23
		UC		0.44	0.47	0.29		0.42	0.20*
	Pain	CD	0.58				0.59
		UC	0.78				0.54
	Tired	CD	0.57		0.05*		0.44		0.33	0.46
		UC	0.62		0.27		0.35		0.30	0.34
	Annoyed	CD		0.15*				0.34	0.21	0.28
		UC		0.42				0.37	0.44	0.29
	School-work	CD			0.17*
		UC			0.38					0.40
	Sleep	CD	0.13*						0.24	0.28
		UC	0.46						0.25	0.17*
	Daily Routine	CD	0.33		0.15*		0.34		0.26	0.39
		UC	0.40		0.30		0.54		0.26	0.25
	Activities	CD	0.50		0.13*		0.60		0.50	0.52
		UC	0.64		0.46		0.62		0.43	0.36
HUI2	Emotion	CD		0.37		0.27		0.49
		UC		0.40		0.41		0.49
	Cognition	CD			0.24				0.39	0.40
		UC			0.13				0.13*	0.35
	Mobility	CD	0.05*				0.34
		UC	0.24				0.30
	Self-care	CD	0.10*				0.27
		UC	0.30				0.36
	Pain	CD	0.64				0.61*
		UC	0.61				0.57
HUI3	Emotion	CD		0.05*		0.39		0.39
		UC		0.46		0.57		0.53
	Cognition	CD			0.22*				0.41	0.42
		UC			0.10				0.17*	0.34
	Ambula-tion	CD	0.00				0.35
		UC	0.24				0.30
	Pain	CD	0.71				0.66
		UC	0.73				0.65

^*Indicates results were not statistically significant, P > 0.05 for the particular correlation tested. Domains not shown were not tested. Bold type indicates correlation coefficients that agreed with the a priori hypothesis. CD= Crohn’s disease; UC= ulcerative colitis.

For PedsQL, as hypothesized, stronger correlations (r > 0.5) were observed between CHU9D Worry and PedsQL Emotional Functioning (r = 0.57), CHU9D Pain and PedsQL Physical Functioning (r = 0.59), and between CHU9D Activities and PedsQL Physical Functioning, Social Functioning and School Functioning domains. In some cases, an expected correlation was not observed such as between CHU9D Schoolwork and PedsQL School Functioning (P = 0.48, P < 0.05). For HUI, HUI2 Pain was moderate-to-strongly correlated with PedsQL Physical Functioning (r = 0.61, P = 0.6), but this was not statistically significant. It was hypothesized that HUI2 and PedsQL Emotion domains would be strongly correlated, however these domains demonstrated r = 0.49. HUI3 Pain and PedsQL Physical Functioning were strongly correlated (r = 0.66, P < 0.05), however, the Emotion domains of HUI3 and PedsQL were not strongly correlated (r = 0.39, P <0.05). For both the HUI2 and HUI3, Cognition domains were, as hypothesized, moderately correlated to PedsQL Social Functioning (r = 0.39 and 0.41, P < 0.05, respectively). It was hypothesized that the Cognition domain in the HUI2 and HUI3 would be strongly correlated to PedsQL School Functioning, however r = 0.40 and 0.42 (P < 0.05), respectively.

Correlation between Preference-Based and Non-Preference-Based HRQOL Measures in UC

In UC, correlations between instruments differed slightly from those observed in CD (Tables 3 and 4). The CHU9D and HUI showed similar moderate to strong statistically significant correlations with IMPACT-III ranging from 0.67 to 0.81 across questionnaires and tariff sets. The CHU9D demonstrated stronger correlations with the PedsQL (range 0.69-0.71) than HUI (range 0.51 to 0.67). As presented in Table 4, more so than observed in CD, CHU9D Pain was moderate-to-strongly correlated with IMPACT-III Well-being (r = 0.78, P < 0.05). Similarly, CHU9D Worry was moderately-to-strongly correlated (r = 0.63, P < 0.05) with IMPACT-III Social Functioning and Emotional Functioning (r = 0.54, P < 0.05). It was expected that CHU9D Sad would be more highly correlated to IMPACT-III Emotional functioning than observed (r = 0.44, P < 0.05). CHU9D Sad was moderately correlated to IMPACT-III Social functioning (r = 0.47, P < 0.05). CHU9D Activities was also moderately correlated with IMPACT-III Social functioning (r = 0.46, P < 0.05). In general, CHU9D correlations with IMPACT-III were higher in UC than in CD. For example, CHU9D Activities was moderately-to-strongly correlated with IMPACT-III Well-being (r = 0.64, P < 0.05), and CHU9D Daily Routine was moderately correlated with IMPACT-III Well-being (r = 0.40, P < 0.05). Similarly to CD, HUI2 Pain was moderate-to-strongly correlated with IMPACT-III Well-being (r = 0.61, P < 0.05). Unlike CD, HUI3 Emotion was unexpectedly more strongly correlated with IMPACT-III Body Image (r = 0.57, P < 0.05). Similarly to CD, HUI3 Pain was moderate-to-strongly correlated with IMPACT-III Well-being (r = 0.73, P < 0.05).

With regard to PedsQL, CHU9D Worry was not as strongly correlated with PedsQL Emotional functioning (r = 0.49, P < 0.05) as was observed in CD. It was expected that CHU9D Sad and PedsQL Emotional functioning would be strongly correlated however, this was not observed in CD or in UC (r = 0.41 and 0.42, P < 0.05, respectively). As in CD, CHU9D Pain was moderate-to-strongly correlated to PedsQL Physical functioning (r = 0.54, P < 0.05). CHU9D Activities was moderate-to-strongly correlated with PedsQL Physical functioning (r = 0.62, P < 0.05), but was less correlated with Social functioning and School functioning compared to CD (r = 0.43 and 0.36 respectively, P < 0.05). The greatest differences between UC and CD were observed in the correlations between CHU9D Daily Routine and PedsQL Physical functioning (r = 0.34 in CD and r = 0.54 in UC, P < 0.05), and CHU9D Annoyed and PedsQL Social functioning (r = 0.21 in CD and r = 0.44 in UC). For HUI2, Emotion was not as strongly correlated with PedsQL Emotional functioning as expected, r = 0.49, P < 0.05. HUI2 Pain was moderate-to-strongly correlated with PedsQL Physical functioning of (r = 0.57, P < 0.05). HUI2 Mobility and Self-care domains were expected to be more strongly correlated with PedsQL Physical functioning than they were observed to be. HUI3 Pain was moderate-to-strongly correlated with PedsQL Physical functioning (r = 0.65, P < 0.05). In contrast to CD, in UC, HUI3 Emotion was moderate-to-strongly correlated with PedsQL Emotional functioning (r = 0.53, P < 0.05).

DISCUSSION

In addition to enabling the comparison of HRQOL between different patient populations, generic preference-based HRQOL questionnaires have the added benefit of enabling the calculation of QALYs—a universal outcome measure—to permit decision makers to compare the value of treatments across diverse patient populations when budgets are constrained. It is therefore essential that generic preference-based questionnaires are administered in observational studies or clinical trials evaluating new treatments. Unfortunately, few instruments exist or have been validated for use in children (16). This study aimed to address this gap for pediatric patients with IBD.

Moderate to strong correlations were observed between preference-based CHU9D and HUI and the disease-specific IMPACT-III or generic PedsQL. The Pain domain in all preference-based questionnaires was most strongly correlated to the Well-being and Physical functioning domains of the IMPACT-III and PedsQL questionnaires, respectively. While most correlations between CHU9D, HUI2 and HUI3 and the IMPACT-III and PedsQL were similar, there were important differences between generic preference-based questionnaires and between CD and UC populations. Overall, the CHU9D with youth tariffs was slightly more correlative to the IMPACT-III and PedsQL than the CHU9D with adult tariffs. Further, the CHU9D showed stronger correlations with the IMPACT-III and PedsQL questionnaires in UC than in CD. In UC, the HUI3 showed a stronger correlation to the IMPACT-III and PedsQL questionnaires than the HUI2.

HRQOL measurement have been previously reviewed in children and in adults with IBD (2,3,17,18) but those reviews focused on disease-specific measures. The most commonly-used disease-specific HRQOL questionnaire in children with IBD is the IMPACT-III (18). No commonly-used generic HRQOL questionnaire was identified in children with IBD, but the SF-36 generic HRQOL measure was frequently used in adults (17,18). When generic HRQOL assessments were compared to IBD disease-specific instruments, Knowles et al (17). concluded that generic HRQOL measures may be most practical in cross-disease comparisons, but they could underestimate IBD impact. Conversely, disease-specific assessments do not capture all HRQOL domains such as hearing, vision, speech and dexterity. These domains may impact patients with IBD who experience extra-intestinal manifestations such as uveitis.

While the generic PedsQL has been used in pediatric IBD (19), health utilities cannot be generated directly from PedsQL scoring, and therefore QALYs cannot be determined. Researchers have attempted to map PedsQL results to the CHU9D through regression modelling (20,21). One such study was conducted in adolescents and may not be generalizable to younger children such as those typically seen in IBD (20). Another study conducted was applicable to those aged 5 years and older (20,21). Mapping studies have been criticized however and the Canadian Agency for Drugs and Technologies in Health (CADTH) and the United Kingdom’s National Institute for Health and Care Excellence (NICE) that evaluate economic evidence specifically discourage the use of mapping to generate health utilities (22,23).

The choice of an appropriate HRQOL questionnaire in children depends on the study aim, however choosing an instrument that could serve multiple objectives such as assessing HRQOL in addition to generating utilities for QALY calculations would lessen the burden on research participants. Points to consider when choosing a HRQOL questionnaire include the performance characteristics such as the validity, reliability and responsiveness of the questionnaire within the patient population of interest, the age group for whom the questionnaire has been validated, the recall period incorporated into the questionnaire, the burden of administering the questionnaire in terms of length, feasibility and follow-up time-points, and the long-term objective of administering the questionnaire. With regard to pediatric IBD, while the IMPACT-III has been specifically developed and validated to assess HRQOL in children with IBD, it cannot be used to generate health utilities. Similarly, the PedsQL can help to compare HRQOL scores across different pediatric diseases and age groups, but does not produce health utilities. The Short Health Scale (SHS) quality of life questionnaire was validated and correlated with the PedsQL and IMPACT-III in children with IBD (24). While this tool is brief and adequately assesses HRQOL in children, it is not preference-based and it too cannot be used directly for determining health utilities. For the purpose of generating utilities to calculate QALYs, the present study suggests that the CHU9D with youth tariffs and the HUI3 may be useful generic preference-based options for utility generation in pediatric IBD. The CHU9D appears to be a reasonable and sensitive measure of pediatric IBD HRQOL. However, as a pediatric-specific tool, the CHU9D would be limited in generating QALYs for lifelong conditions such as IBD where utilities for adult health states are also needed. In contrast, the HUI can be used for all persons over 6 years of age. Until HTA agencies broaden their acceptance of methodologies used to elicit health preferences, the HUI3 may be the most suitable generic preference-based questionnaire to elicit health utilities in children and adults with lifelong conditions such as UC and CD.

While a strength of this study was comparing different pediatric utility instruments within the same groups of patients, the study did not include additional measures such as the EQ-5DY, a youth version of the EQ-5D (25), or direct approaches to utility elicitation, such as standard gamble, time trade-off or best-worst scaling. These cognitively complex direct approaches can be challenging to administer in children. Another strength was the comparison of performance of measures between CD and UC pediatric populations derived from data collected at the same clinical sites, thereby controlling for variations in practice patterns or standards of care. However, enrollment was limited to three Canadian centres which may limit generalizability. The need to examine correlations between instruments by ethnic diversity and gender is growing in importance to expose inequities in health outcomes. This study assumed that given the similarities in population demographics, health care systems and treatment of IBD, Australian youth tariffs would be applicable to Canadian children with IBD. Previous work comparing the CHU-9D to the HUI which used Canadian tariffs showed its applicability to the Canadian population (7).

CONCLUSION

This study was undertaken to generate evidence on the optimal means of generating health utilities for use in economic evaluations of treatments in pediatric IBD. The analysis found that generic-preference-based questionnaires was moderately to strongly correlated to with conceptually overlapping domains of the IMPACT-III and PedsQL. Correlations differed slightly between UC and CD populations. While all questionnaires were moderately correlated with the IMPACT-III and PedsQL questionnaires, overall, the CHU9D using youth tariffs and the HUI3 exhibited the strongest performance for use in children with IBD.

Author Contributions

Naazish Bashir was involved in the study concept and design, acquisition of data, statistical analysis and interpretation of data, drafting of the manuscript, critical revision of the manuscript for important intellectual content, approval of the final manuscript, and obtaining funding support. Thomas Walters was involved in study supervision, study concept and design, acquisition of data, interpretation of data, critical revision of the manuscript for important intellectual content, and approval of the final manuscript. Anne Griffiths was involved in study supervision, study concept and design, interpretation of data, critical revision of the manuscript for important intellectual content, and approval of the final manuscript. Anthony Otley and Jeff Critch were involved in study supervision and critical revision of the manuscript for important intellectual content and approval of the final manuscript. Wendy Ungar was involved in study supervision, study concept and design, acquisition of data, interpretation of data, critical revision of the manuscript for important intellectual content, approval of the final manuscript, and obtaining funding support.

Conflict of Interest

Naazish Bashir has no conflicts of interest. Thomas Walters has served on advisory boards for AbbVie and Janssen, received research support from AbbVie, and participated in speaker’s bureaus for AbbVie, Janssen and Nestle Health Sciences. Anne Griffiths has been a consultant for Abbvie, Amgen, Bristol Myers Squibb, Janssen, Lilly and Merck have received speaker fees from Abbvie, Janssen and Nestle, and research support from Abbvie. Anthony Otley has been a consultant for AbbVie, an advisory board member for AbbVie and Janssen, and has received research support from AbbVie, Lilly, Takeda and Janssen. Jeff Critch has been a consultant for Abbvie and Pharma Science. Wendy Ungar has no conflicts of interest.

Data Availability

The raw data underlying this article cannot be shared publicly as participants did not provide consent for data sharing. The raw datasets generated during and/or analyzed during the current study are proprietary to The Hospital for Sick Children. Data summaries are available at https://cidscann.ca.